Induced chemical defenses in invasive plants: a case study with Cynoglossum officinale L.
- First Online:
- Cite this article as:
- Eigenbrode, S.D., Andreas, J.E., Cripps, M.G. et al. Biol Invasions (2008) 10: 1373. doi:10.1007/s10530-007-9212-z
- 197 Views
The ‘evolution of increased competitive ability’ (EICA) hypothesis is an extension of optimal defense theory and predicts that reduced pressure from insect herbivores in the introduced range results in evolution of reduced defenses in invading plant populations, allowing greater allocation of resources to competitive traits such as growth rate and reproduction. The EICA hypothesis considered levels of defensive chemistry to be fixed within a particular genotype. In this paper, we propose that if herbivory is reduced in the introduced range, but chemical defenses are inducible in response to herbivory, evolution of reduced defenses and any associated increase competitive ability should not occur. Rather, mean induced and constitutive levels of induced defenses should be similar in introduced and native ranges, but the variance about mean induced levels should be greater in the introduced range. This is predicted because induced levels will occur less frequently in the introduced range where herbivory is reduced, thereby insulating these levels from the stabilizing selection expected in the native range where induced levels occur more frequently. We conducted a preliminary study to examine this by comparing constitutive and induced concentrations of total pyrrolizidine alkaloids (PAs) from native (European) and introduced (western North America) populations of Cynoglossum officinale L. The mean constitutive and induced concentrations of PAs did not differ between continents, but the variability of the induced concentrations was significantly greater for plants from the introduced range. Although our study with C. officinale is provisional due to a small sample size, it supports our predictions for evolution of inducible defenses in introduced ranges where herbivore pressure is reduced. Most chemical defenses in plants have been found to be inducible, so similar patterns may occur widely. If so, this weakens the generality of EICA’s predictions concerning chemical defenses. The effects of inducible defenses should be considered in cross-continent comparisons of other invasive plant species.
KeywordsEICAHoundstongueHerbivore defenseHerbivore resistanceInduced defenseInvasive species
The ecological and economic effects of exotic plant species are alarming (Mooney and Hobbs 2000; Pimentel 2002), but the mechanisms by which some introduced plants become invasive weeds are still poorly understood (Mack 2003; Hinz and Schwarzlaender 2004; Hierro et al. 2005). Both evolutionary and ecological mechanisms have been proposed. The ‘enemy release hypothesis’ (ERH) posits an ecological mechanism wherein plant species become invasive because of reduced pressure from insect herbivores in their introduced ranges (Elton 1958; Maron and Vilà 2001; Keane and Crawley 2002). An evolutionary corollary of ERH, the ‘evolution of increased competitive ability’ (EICA) proposes that relaxation of selective pressure from natural enemies in the introduced range favors evolution of reduced defenses, with consequently greater allocation of resources to competitively advantageous traits such as rates of growth and reproduction (Blossey and Nötzold 1995). The EICA hypotheses has been very influential in determining the research agenda on the mechanisms of invasion (Bossdorf et al. 2005).
The EICA hypothesis and its parent, optimal defense theory (ODT), require that plant chemical defenses exact metabolic or ecological costs such that natural selection will constrain levels of these defenses in proportion to such costs, and the benefits they confer (Herms and Mattson 1992; Strauss et al. 2002; Stamp 2003). Although such costs are sometimes difficult to detect, their widespread occurrence appears indisputable. Costs and benefits of defenses must vary with their modes of action and levels required for effectiveness, the degree of tolerance present in the plants, and ecological factors (Feeny 1976; Rhoades and Cates 1976; Baldwin 1998; Strauss et al. 2002; Stamp 2003; Müller-Schärer et al. 2004).
ODT can be extended to include inducible chemical defenses, which can be viewed as adaptations to prevent allocation of resources to costly defenses until they are required (Karban and Baldwin 1997; Agrawal 2005). The low levels of defenses before induction should have negligible costs and not necessarily be optimized, but the higher, induced levels could be optimized since they will exact metabolic costs and provide ecological benefits that will exert balancing or stabilizing selection on their levels.
We are not aware of studies addressing the implications of inducible defenses for the predictions of plant defenses in invaded ranges. One study examined patterns of inducible defenses in native and introduced ranges in the context of EICA (Cipollini et al. 2005) but did not find consistent patterns among the specific defenses examined, nor did the authors formulate predictions for how induced defenses could influence the predictions of EICA. Some predictions can be offered based on the assumptions that: (1) constitutive levels of inducible chemical defense exact lower or negligible costs as compared with induced levels, (2) herbivory capable of inducing defenses is relaxed in the introduced range, (3) induced levels are effective defenses but are constrained by natural selection due to allocation or ecological costs, and (4) introduced populations possess genetic variability for constitutive and inducible levels of chemical defense.
Where these four assumptions hold, reduced constitutive levels are not expected in the invaded range because their costs are negligible. Induced levels, although potentially costly, will be rarely expressed in the introduced range where herbivory is relaxed, so they will be insulated from selection and are not expected to differ from those in the native range. By contrast, in the native range where induction is more frequent, induced levels should be optimized based on their costs and benefits as predicted by ODT. The reduced frequency of expression of induced levels of defenses in the introduced range, however, should release them from stabilizing selection, making them more likely to vary within and among populations due to genetic drift, founder effects and mutation.
We conducted a preliminary test of these predictions by comparing the constitutive and injury induced concentrations of pyrrolizidine alkaloids (PAs) in populations of Cynoglossum officinale L. collected from its native range in Europe and from its introduced range in western North America. In C. officinale, PAs are inducible by mechanical injury (Van Dam and Bhairo-Marché 1992) and their inducibility is genetically variable (Van Dam et al. 1996), making the species appropriate for examining changes in inducibility associated with introduction.
Cynoglossum officinale (Boraginaceae) is native to temperate Eurasia, and was introduced to North America in the late 1800s as contamination in cereal crop seed (Brand 1921; Tutin et al. 1972). It now occurs from the east to west coasts of North America, but it is particularly problematic in the west where it is declared noxious in six states and two provinces (USDA-NRCS 2005; Rice 2005). In the native range in Europe at least 69 arthropod herbivores are recorded from C. officinale (Freese 1987; De Jong et al. 1990), of which five were considered specialized feeders with potential for classical biological control (De Clerck-Floate and Schwarzlaender 2002). One specialized biological control agent, the root-feeding weevil, Mogulones cruciger, was released in Canada in 1997. The herbivore community on C. officinale has not been investigated formally but invasive populations in North America appear to experience less herbivory than native populations in Europe (MS and JA, unpublished observations).
Collection sites for C. officinale populations
N 47°48′; E 7°36′
N 49°00′; E 8°24′
N 51°30′; E 0°07′
N 47°82′; E 16°24′
N 49°28′; W 114°11′
N 49°33′; W 115°32′
N 46°82′; W 116°42′
Pyrrolizidine alkaloid analysis
Constitutive levels and induced PA levels were measured in individual leaves of 10 plants from each site of origin in mid July 2001. Sampling for constitutive levels was accomplished by removing for analysis one half of a single leaf blade, excluding the midvein, from the 4th node of the plant. Removal of this half leaf constituted the mechanical injury for induction. After 48 h, the remaining half leaf, also excluding the midvein, was removed and assayed for induced PA concentrations. Induction by controlled mechanical damage was used because it is known to be effective in C. officinale (Van Dam et al.1993; Van Dam and Bhairo-Marché 1992) and because our goal was to measure maximum potential inducible levels in the plants. Total PA concentration was determined using the extraction method of Hartmann and Zimmer (1986) followed by colorimetric determination of total PAs (Mattocks 1967) as described in Van Dam and Bhairo-Marché (1992). A monocrotaline standard curve was run with each group of samples. Quantification (mg/g) was based on a monocrotaline standard curve and the dry wt of the plant material.
An analysis of variance was used to assess the effects of continent and collection site (nested within continent) on the constitutive and induced levels of PA. Further analyses (folded F tests) was carried out to compare statistically the variance in PA levels between continents for each treatment. Analyses were conducted in SAS (SAS Institute, Cary, NC) using PROC GLM for ANOVAs and PROC TTEST to compare residuals and calculate folded F tests to compare variances.
Summary statistics from ANOVAs to test for the effects of continent and site on constitutive and induced levels of total pyrrolizidine alkaloids (PA) in C. officinale and t-tests to compare residuals about the means within continent
A. Constitutive PA concentrations
Pr > F
Test of the effect of continent using site(continent) as the error term
Pr > F
Folded F test for equality of residual variances
B. Induced PA concentrations
Pr > F
Test of the effect of continent using site(continent) as the error term
Pr > F
Folded F test for equality of residual variances
If herbivory by generalists or specialists is reduced in the introduced range of a plant species, ODT predicts evolution of reduced defenses in the introduced populations. This prediction is a cornerstone of the EICA hypothesis. If chemical defenses are inducible, however, and only produced in high concentrations in response to herbivory, selection for inducibility and stabilizing selection on the induced concentrations of defensive chemicals will be reduced if herbivory is reduced in the introduced range. The expected outcome will be greater variability in these traits due to drift, founder effects, mutation or some combination of these processes, but not necessarily lower induced or constitutive concentrations of defenses in the introduced range.
Our study with PAs in C. officinale is consistent with these predicted patterns. The relatively low constitutive PA concentrations may not exact metabolic costs sufficient to influence their natural selection. Indeed, the costs associated with defensive PAs in C. officinale have been difficult to quantify (Van Dam and Vrieling 1994; Van Dam et al. 1996) and other studies have failed to detect differences in constitutive PAs from native and introduced ranges of Senecio jacobaea (Stastny et al. 2005; Joshi and Vrieling, 2005).
Induced PA concentrations are at least 4-fold greater than constitutive levels (this study and Van Dam and Bhairo-Marché 1992 and Van Dam et al. 1993) and could exact significant metabolic costs. Costs of producing PAs have been demonstrated in S. jacobaea (Vrieling and Van Wijk 1994) and there are costs implicated in association with transport and reallocation of PAs within C. officinale (Van Dam et al. 1996). Nonetheless, no difference was found in mean inducible levels of PAs between the C. officinale from the two continents. Either the costs of induced levels are insufficient to drive selection, or as we hypothesize, induced concentrations rarely occur in the introduced range because of the reduced herbivory there, and so are little affected there by natural selection. The significantly higher variances in induced PA concentrations in the introduced range (NA) as compared with the native range (EU) is consistent with predictions if induced concentrations were less exposed to natural selection in North America. In contrast, the lower variance in induced PA concentrations in C. officinale from its native range could reflect the action of stabilizing selection on those concentrations constraining them towards optimal levels.
Many plant defenses are inducible by injury and herbivory, so patterns of concentration and variability of chemical defenses in plant populations in native and introduced ranges may follow patterns similar to those observed here for C. officinale. This possibility should be investigated. One other report of inducible defenses compared among continents has found variable results depending on the specific defense (Cipollini et al. 2005) and did not compare variance of defenses between native and introduced ranges. The implications of our study are: (1) inducible defenses can obscure or prevent one of the key predictions of EICA as it was originally formulated (Blossey and Notzold 1995), (2) studies to detect evolution of plant defenses in introduced ranges should include comparisons of induced and constitutive levels of defense, and (3) greater variability in levels of induced defenses in the introduced ranges could cause variable susceptibility of introduced populations to biological control agents.
Phenotypic plasticity can have substantial effects on evolution in a community context by altering the strength and direction of natural selection involving interacting species (Fordyce 2006). Inducible chemical defenses in plants are a widespread type of phenotypic plasticity that appears adaptive and may be fine tuned to the levels and frequency of herbivory experienced by plant populations (Karban and Baldwin 1997). Plant populations that have established new ranges where herbivory is substantially reduced or altered present opportunities to study the evolution of adaptive phenotypic plasticity. Further studies should assess more populations and document herbivore loads and induction status of plants in the field in introduced and native ranges.
The authors thank Jessica Spafford for assistance with data collection and Willliam Price for statistical consultation. Helpful comments on the manuscript were provided by G. Newcomb, J. McCaffrey, and H. Hinz. This work was supported by grants awarded to M.S. and S.D.E. and by the Idaho Agricultural Experiment Station.